The overall objective of the proposed research is to develop effective new antitumor agent. This will be achieved by executing highly stereocontrolled total syntheses of natural products with potent antitumor activity. The syntheses will provide: (1) means for preparing modified derivatives, not available from biological sources, which would exhibit higher therapeutic ratios and selectivities than the natural products, and (2) ready access to radiolableled derivatives which are valuable for studying the binding of such molecules with biological macromolecules as well as the transport and metabolism of these new drugs. We invented a stereospecific rearrangement which produces allylic diepoxides in high yield. An efficient total synthesis of spatol exploiting this remarkable reaction will soon be completed. Our next synthetic objective is halichondrin B, another natural product of marine origin. Spatol inhibits cell division in human T242 Melanoma and 224C Astrocytoma neoplastic cell lines, and the synchronous cell division of the fertilized sea urchin egg. Therefore, this novel diterpenoid diepoxide will be a useful template for the design of chemotherapeutic agents for, inter alia, human skin and brain tumor inhibitors. The management and cure of brain and other central nervous system cancer by chemotherapy are very urgently needed. Hopefully, with the use of spatol or analogues crossing the blood-brain barrier, this most worthwhile goal will be realized. Radiolabeled spatol will now be prepared and used to study its antimitotic action. Interactions with microtubule proteins will be characterized to compare the site of action with that of several other mitotic spindle poisons, such as colchicine, maytansine, vincristine, and vinblastine, which have proven clinically useful. Halichondrin B possess extremely potent antitumor activity against B-16 melanoma, as well as P-388 and L-1210 leukemias in vivo. Since more than four billion stereoisomers of halichondrin B are possible, stereocontrol must be a major objective of any total synthesis. Our synthetic strategy benefits from the high efficiency of convergence, and the rich stereochemical and functional content of readily available sugars, D-glucose, D- mannose, and D-ribose, which will be our starting materials. The use of a trimethylsilyl group as a stereocontrol element during epoxidation of homoallylic alcohols is suggested, and new methods are proposed for achieving stereoselective homologation of sugar aldehydes.